SA516371895B1 - Flow Regime Identification of Multiphase Flows by Face Recognition and Bayesian Classification - Google Patents

Abstract

Multiphase flow regimes in downhole or surface flow lines are made identifiable by forming images of the flow based on data obtained by flow regime metering of the flow. The flow regime metering data are gathered and processed in a format in which it can be subjected to face recognition processing of the type used for recognition of the faces of persons, and also to Bayesian classification techniques. Identification capabilities are particularly enhanced in flow regimes where gas bubbles or large amounts of free gas cause clutter and multiple reflections in the metering data. Fig. 1

Description

Flow Regime Identification of Multiphase Flows by Face Recognition and Bayesian Classification FULL DESCRIPTION BACKGROUND OF THE INVENTION The present invention relates to the imaging of multiphase flow in channels and is more specific By defining a flow system by virtual distinction and hypothetical classification of multiphase flows of oil, water (brine) and gas in flow conduits. The technique generally used to recreate two phase streams is based on what is known to be a filtered back projection algorithm This type of flow reconstruction is described for example by Kak,

Avinash C., Slaney, Malcolm “Principles of Computerized Fluid transit 0

Imaging,” IEEE Press, New York, USA (1988), and Murrell, H. “Computer-Aided Tomography,” The Mathematical J. V6 (1996), pp. 60- .65; gas and oil present when producing oil fluids though; Because of the nature of the fluids, it is necessary to form images of a three-phase flow in channels that were included when producing hydrocarbons. Because of the different fluid properties of water (brine, oil and gas), it is difficult to treat all three groups of fluids. At the same time In the case of multiphase oil-water or water-oil flows the medium is used In the case of liquid-gas or gas-liquid flows (where the liquid is the brine, oil or

both of them); The attenuation method was used. as it's known; Nevertheless; Neither method provides a fully satisfactory determination or temporality of a cross-section of a three-phase polycyclic flow. One type of flow regime where it is observed that at least a fine line occurs when there are elevated levels of free gas within the conduit system. These levels of gas cause all 985 766 -multiple reflections Cua tomography generally depends on transmission other than more reflections ages can result from the presence of gas bubbles or large amounts of All gas has complex turbulence and reflections that impede when defining a flow regime based on a tomographic reconstruction. Determining the flow regime in canal systems also proves difficult as there are conditions of 0 gal measurement which represent ie hans moist gas or saturated flow into which the bubbles have been bubbly pumped. GENERAL DESCRIPTION OF THE INVENTION IN BRIEF The present invention provides a new and improved device for determining flow regimes for multiphase flow in a conduit. The device includes an array of ultrasonic transceivers that are installed around the perimeter of the channel transceiver energy to travel through the fluid in the channel. The group of 5 transceivers which are installed around the perimeter of the channel also receive the energy after traveling through the fluid in the channel. The instrument includes a dallas data processing system that modulates the flow system signals for the multiphase flow in the channel. The memory of the data processing system stores a database of flow system data as a set of data matrixes representing the test flow conditions in the channels for the group of test flow systems in the channel. The 0 data processing system also includes a processor that forms data metrics for the fluid transfer of energy between individual ding transceivers of the transceiver group as a result of the transmission through the fluid in the channel. The processor also organizes the fluid transmission data metrics into the 76 data matrixes which indicate the actual flow conditions in the channel. The processor forms the AIS correspondence measure between the data in the actual flow conditions in the actual data matrix and the individual 25 1007/08 matrixes from the set of data matrices that Jia test flow conditions in

Channels for a range of test flow systems. The processor defines the flow regime for the actual flow conditions in the channel based on the metrics configured for symmetry. A demonstration of a data processing system provides a flow regime specified for actual flow conditions for evaluation and analysis. The present invention also provides a new and improved data processing system for determining flow regimes for a multiphase fluid flow in a cle channel on the basis of energy transmission through the fluid in the channel from a transmitter A receiver to transmit to a group of transceivers around the perimeter of the channel to a group of transceivers to group receivers. The DPS memory stores a database of flow system data as a set of data matrices representing test flow conditions in the channels for a group of test flow systems in the channel. The data processing system also includes 0 processor that forms data metrics for fluid transmission of energy between individual transceivers and the rest of the transceiver group as a result of transmission through the fluid in the channel. The processor also organizes the fluid transmission data metrics into the actual data matrix indicative of the actual flow conditions in the conduit. The processor also modulates the correspondence measure between the data in the actual flow conditions in the actual data matrix and an individual set of groups of data matrices representing the test flow conditions 5 into the channels for a range of test flow systems. The processor defines a flow regime for the actual flow conditions in the channel based on the metrics configured for symmetry. The data processing system display presents the specific flow system of actual flow conditions for evaluation and analysis. The present invention also provides a new and improved computer-implemented method for constructing flow system images of a three-phase flow in a conduit based on the transmission of energy through the fluid in a conduit from a transmitter 0 to a set of transceivers installed around the circumference of the conduit to a set of receivers in an array from the transceiver devices. A database of flow system data is stored as a set of data matrices representing test flow conditions in the channels for a group of test flow systems in the channel. Fluid energy transfer data metrics between individual devices of the dig group transceivers were formed

Transceivers as a result of transmission through the fluid in the channel. The fluid transport data metrics are organized into an actual data matrix indicative of the actual flow conditions in the conduit. A measure of correspondence between data in actual flow conditions is formed in the actual data matrix and an individual matrix of a set of data matrices representing test flow conditions in channels for a range of flow systems

the test. The flow regime is selected for the actual flow conditions in the channel on the basis of the parameters formed for symmetry; ang Provide specific flow system of actual flow conditions for evaluation and analysis. The present invention also provides a new and improved data storage method stored in computer-operable instructions for a non-transition computer-readable medium to cause a data processing system to form flow system images of a three-phase flow in a channel on the basis of energy transfer

0 through the fluid in a channel from a transmitter from a group of transceivers installed around the perimeter of the channel to a group of receivers in a group of transceivers. The instructions stored in the data storage medium cause the data processing system to store a database of flow system data as a set of data matrices representing the test flow conditions in the channels assigned to a set of test flow systems in the channel.

5 The instruction causes the data-processing system to construct metrics of the fluid transfer of energy data between individual transceivers and the rest of a group of transceivers as a result of the transmission through the fluid in the channel; and to organize the fluid transmission data metrics into the actual data matrix indicative of the actual flow conditions in the channel. The instructions stored in the data storage medium cause the data processing system to form a measure of correspondence between the data in the actual flow conditions

0 in the actual data matrix and an individual set of data matrices that Jia test flow conditions in channels for a group of test flow systems. The instructions stored in the data storage medium cause the data processing system to determine the flow regime of the actual flow conditions in the channel on the basis of the formed symmetry metrics and advance the flow system defined of the actual flow conditions for evaluation and analysis.

Brief explanation of the drawings Figure 1 is a stereoscopic perspective view; Partially formed in the form of a schematic form; of an ultrasonic imaging system that is fixed using a conduit according to the present invention. Figure 2 is a vertical cross-sectional view of a fluid flow through the sonographic system of Figure 1. Figure 3 is a schematic of the data acquisition sequence in the sonographic system of Figure 1. Figures 14 and b are flow-edge diagrams for two of the systems The multiphase flow formed according to the present invention. 0 FIGURE 5 is a functional frame diagram of a set of steps performed according to the present invention in the preprocessing of the flow system data for processing according to the treatment of Fig. 8. Figs 6 and 6b are diagrams of the flow system data present during the processing according to Fig. 5. Fig. 7 is a statement A functional framework diagram of a set of steps performed according to the present invention 5 when creating a database of flow faces representing multiphase fluid flow systems during the processing shown schematically in Figure 8. Figure 8 is a functional framework diagram of a set of steps performed when acquiring Flow system data for multiphase flow imaging performed according to the present invention. Fig. 9 is a schematic of the data processing system to determine the flow regime of multiphase 0 flows by means of virtual discrimination and a hypothetical classification according to the present invention.

Detailed Description: Flow System Data Acquisition: According to the present invention; Fluid transport data are acquired through ultrasonic measurements by the tomographic reconstruction data acquisition method pursuant to the US patent application which has fallen into the public domain and is pending by Applicant No. 595689/14. The present invention then organizes the acquired fluid transport data measurements to enter into

Virtual discrimination data processing. after that; The present invention uses machine learning and hypothetical classification techniques around this data. Thus the invention all makes use of the whole set of data; Unlike tomographic reconstruction. There is an enrichment of the received data in complex flow inversions in a liquid gas to hinder reconstitution performance

0 tomography. using the present invention; The behavior and patterns of complex flow data are useful for determining the exact flow regime. Referring to Figure 1; A measurement system for the ultrasonic flow system “TA” for data acquisition from device “A” according to the present invention is illustrated to determine the flow regime for a three-phase fluid flow in a channel of a ‘C©’. Device “A” also includes a data processing system "d"

5 (Fig. 9) and a method (Figs. 5; ¢7 and 8) provide determination of the flow regime for multiple “lela” mixtures as will be described below. In this preferred embodiment, the three phases represent oil, water (brine) and gas based on measurements from the system. The “T1” flow. The “T” flow system of the present invention takes the form of the ‘Ula’ array of ultrasonic transceivers 20 which send the ultrasonic energy to travel through the “C” flow channel.

For example, the set of production tubing in a wellbore or the AT pipe through which the flow is or can occur in a multiphase. It should be understood that other channels, both downhole and at the surface, through which the flow of the three-phase fluid is measured Phases (water (or brine); oil and gas) can represent the subject of flow regime identification from flow system data acquired according to

of the present invention. A standard design is taken into consideration for the measurement of the ultrasonic flow system as shown in Figure 2. The measurement system ‘T’ of Figure 2 shows a set of ‘N’ transducers positioned in a circular ‘U’ matrix evenly spaced around the outside of the The production pipe or assembly 24 of the "C" flow conduit having a polyphase flow pattern possibly comprising a continuous asphalt phase 30; a gas phase 32; or a water phase 34. The exact design of the polyphase flow can be varied for different aS distribution deviating than the one shown in Figure 1, but here it is for illustrative purposes When measuring, each of the N sensors pulses in turn Monitoring the signals from the remaining sensors (N-81-11) for a period of time starting from the start of the pulse 0 until such time as all the sound field inside the tube stops. For 3" diameter tube using transformers operating at 330 kHz; This timescale is on the order of 200 microseconds. The transmission of energy through the fluids takes place in the 'C' channel through the network of transmission channels schematically indicated at 22. This results in a set of transmission paths N (N-1) 22 that can be seen 5 (Silas for the laying of the pipe strings that have been Configured when playing the cat's cradle The fluid properties can be measured along individual channels of the 22 transmitters as represented as a line passing from transmitter 20 to receiver 20. The transceivers 20 are installed in a channel “c” ¢ ie Length of the output tubeset 24. The transceiver 20 is acoustically closely coupled to the polyphasic flow schematically indicated ad) by the arrows 26 passing through 0 of the tubeset 24. In general; There are transceivers 'N 1! ' 20. Figure 3 shows sixteen as an example using all letter-numbered 20 through 20p, respectively. To make the measurement, a single transducer is selected, ie, as indicated at the sensor. sensor 20 in Fig. 2 to be an illustrative emission source The remaining sensors (20b through 20p; 5 in order) are selected as receivers In an illustrative embodiment of the present invention the transceiver devices are 20

At each position 120 through 20p shown in Figure 2 are 333 kHz ultrasonic transducers manufactured by Pro-Wave in Taiwan. The pulse produces a burst of ultrasonic energy that propagates outward from the transceiver 20a. for each receiving sensor 20b up to 0; Latency and attenuation can be measured as fluid travel data and transmitted to the D data processing system (Fig. 9) for storage and sequential processing; It will also be described. It should be noted that non-uniform correction is required to measure attenuation because the gain from each sensor A through Z can be different. This non-uniform correction can be calculated initially for the flow regime T data acquisition system prior to diffusion within a known homogeneous fluid; like; slo distilled into a test channel or container. 0 Figure 2 shows N-1 transmission paths produced from a 20A transmitter. in bubble infusion or gaseous sites; Multiple reflections or refractions of sound waves can result in multiple, complex transmission paths from transmitter to receiver. In Figure 2 if there is an even number "N" of transceivers; The increment of the angle © between adjacent paths (eg » angle I-A-H) is equal up to N/T = 0 radial angle. Taking the line a 5 - a reference angle 0 (regardless of the angular orientation of the source); The ultrasonic transmission data in the form of latency and attenuation measurements produce a set of n = 1 measurements with the following angles measured with respect to angle 0 m representing a central measurement of 714 that is a straight diagonal through the tube.

N 7 7 Nn = [= (5 — n) and h Nz 6 wherel<n<N-1 (1) Equation (1) applies only to even values of if. Directly, odd values are computed, but It cannot generally be taken into account when measuring because it does not provide a straight cross measurement with I = 0. For each of the 'N' transceivers (there are N-1 measurements of the transceiver voltage AS time that can be; For example; between 200 to 2000 or metrics

— 0 1 — over voltage within a time frame of 0 - 500 µs for 3" tube of flux channel system. Time to pulse initiation measured on transmitter; fia 20a. The transmitted pulse is typically a square voltage pulse of 10-100 V half the duration of the excitation frequency of the transceivers - in the case of the 330 kHz means used; the pulse duration is 1-2 microseconds).

The number of samples in each time measurement is specified as an “S” representing the number of samples per channel; that can range»; for example » from about 200 to about 5000. With measurements taken farther out on longer time scales; like; Those that allow multiple paths of ultrasound through a multiphasic medium and can allow for further differentiation. Data acquisition is indicated

0 Ultrasonic flow transmission by group “U” in the manner described LEF schematically at step 30 in the “F” process flow diagram (Fig. For a measurement pulse of 1 for a single ¢ a matrix is defined with dimensions of N-1 X s where each row of the matrix represents an individual dad of AT in descending order and each column of the matrix represents a sampling point

5 individual times in one of each of the sampled N-1 channels. This matrix is defined as follows:

(11070 wherel<m<N-1

Each pulse to a different transceiver produces a new value of (00) a/a. For clas data collection two options can be implemented according to the present invention:

0 Option (1): Concatenate or link together on the image of the JK group [109 101 MD Matrix N- XS selected Majlis bh Option (2): Assessment Procedure: 0/07 1901-1 - ”

A - Al

Regardless of whether Option (1) or Option (2) is used; The same angular offset is used. The benefit of using option (1) is that the data is not discarded or Baha is indifferent while giving dea ll a better rating down the line; But at the cost of requiring increased processing power because the arrays are larger. The benefit of using option (2) is that smaller data sets are required to compute less and averaging can mitigate some of the signal noise.

Nevertheless; Some flow system information is eliminated by the mediation process. According to the present invention; It turns out that the flow system data obtained from ultrasonic flow system measurements using the set 'U' that were organized into either of the two matrices both M' and M' and processed according to the method described below represent what can be determined as 'dnl flow'. He is

0 of the tests obtained values for lm" in the water-air flow loop and indicate the ability to visually differentiate between bubble flow and continuous water flow when presented as Jie these "flow facets". They are illustrated in Figures 4a and e. 14 m-value images formed according to the present invention of continuous water flow using an ultrasonic array of the N-16 channel for measurements. The shape represents an image of al values formed using the same ultrasonic set of flow

5 continuous sparse bubbles. It is clear from Figures 14 and 4b that it is possible to differentiate via the flow system by visual inspection of these images. in Figure 4a; It is observed that the clearly defined crescent shape that indicates transmission through a homogeneous liquid medium at the same speed of sound. The extension or compression of the crescent in the x-axis indicates an increase or decrease in the speed of the sound of the medium

0 . The brightness or darkening of the crescent intensity will indicate a relevant decrease or increase in the attenuation of the medium as it represents an unmeasured signal. The presence of transmission reduction or noise in the medium as shown in Fig. 4b Va indicates the presence of reflective components; like; There are bubbles in the middle. In simpler terms the search for variance may be related to either a local or a global picture of the x-axis extension or the skew of the crescent; darkening or

signal brightness; or for the presence of reflectional noise in the presence of variations in the speed of sound; attenuation or reflective sources, the search will result in different multiphase flow patterns as an example” and the presence of bubble gas flow will result in the presence of multiple reflection sources that result in noise on the signal as shown in figure deg ocd 5 for example; multiphase liquid scenario; like; Bitumen and water produce die

contraction across the crescent shape without adding clutter using the present invention; The benefit of organizing the data in this way is that images can be inserted into Bil methods using image dallas and recognition techniques of the kind used to identify human faces; and default classification techniques to place images into categories between which the possibility of defect is minimized

0 is statistically ranked as the minimum. Once processed using the "D" data processing system (Fig. 9) in this way; A pipe or conduit "C" flow system is made identifiable as a result of its comparability with flow systems of known characteristics determined under conditioning test loop conditions and shared as a flow system database. Data processing: A comprehensive method for determining the ay flow system of the invention Mall is used is illustrated

Ultrasonic measurement data obtained from flow channels using the “T” measurement system schematically in the “F” flowchart (Fig. 4). The "F]" flowchart illustrates the logic structure of the invention as embodied in the computer program. Those skilled in this field will recognize that process flow diagrams show the structures of computer program code elements including logic channels on an integrated circuit that operate according to

0 of the present invention. and clearly; The invention is exercised in its basic embodiment by using a machine component that renders elements of program code in a form representing a digital processing device (ie, a computer) to perform the data transformation sequence or processing steps corresponding to those illustrated. To obtain fluid transport data (either in a flow system database generated or in known flow system conditions) in an acceptable form for computer processing in a D data processing system

(Fig. 9); It is preferable to perform the initial signal processing and measurement in a sang 0 signal processing module (Fig. 5). Figure 16 shows the illustrative raw fluid transport data from a single channel measurement of the U group before treatment; Figure 6b shows this data after processing in a signal processing module 40. In a signal processing module 40; The shal performs a set of steps as schematically shown in Figure 5. The processing to be performed can be carried out in a processing module 40; for example» in a device using» for example; FPGA circuit (or ALE field programmable gate array/A8178 (field programmable gate as ad) at 42; multi-collision FFT filtering is performed on channel fluid transmission data to remove direct current 0 or dal) d.c. ; Low frequency atrophy and interfering signals. as indicated a) at 44; A demodulation filter is used to pass low frequencies over the filtered data generated from Step 42 to provide an envelope response for high-frequency modulations; Compared to demodulating the built-in capacitance or radio signal. as indicated at 46; The resulting spectrum from the demodulation filter in step 44 and a peak or maximum amplitude set equal to 5 13) is scaled using lower amplitudes scaled proportionally as the fractional value of the peak amplitude. After processing in moduleddads 40 The fluid transport data is prepared in matrix form such that the raw flow system data are available as flow faces in the “m” or “al” matrices for further processing when the “d” data is processed according to the present invention, as shown schematically 0 in Fig. 7. During This processing, as will be shown, prepares the data of the stream faces in the m" or al matrices for image identification processing. The stream dng data is taken as Jia image or 0070 image quantity. Accordingly, the flow system data in Each of the matrices m" or m"" for each dag or stream image is a harmonized set of stream images [07); average @ as:

— 4 1 — — 1 = 06m ner ~= (n) Each image stream in the arrays also has what is called a pV image representing the distance of that image from the mean that is specified as toh pm - op = is Clarify further details for determining the image magnitudes' mean or the median values and images on

Sabil Al-Mattal in Sirovich et al., “Low-dimensional procedure for the characterization of human faces”, Journal of the Optical Society of Vol. 4, p. 519 — 524, March 1987, 0781108. after that; The flow image data can optimally be processed with what are known as Igne flow facets according to the present invention.

0 Creating a database of flow faces representing flow conditions: For the determination of the flow regime by image recognition according to the present invention; The construction of a consistent flow system set-up of a flow system or a database of a representative flow facet is formed first as indicated at 70. Using the invention (Jad) the recognition of a flow system is based on the data ALL obtained in the flow loop other than a water squirt as it is Easy to create different model flow systems.Production and data acquisition should be done

5 Flow Loop Knowing that the model flow regimes should represent the flow regimes within i. The ALL or domain behavior can be mapped to the behavior of the flow loop by carefully considering the following physical properties or factors relating to the three fluid phases: (a) maximum and minimum flow rates of the oil; (b) the maximum and minimum flow rates of water or brine;

0 (c) maximum and minimum gas flow rates; (d) Viscosity and density of oil ¢ (e) Viscosity and density of water or brine;

— 5 1 — (f) viscosity density of gas jlall 43USy ¢ (g) for the resulting mixture viscosities and densities using the appropriate mixture correlations. Figure 7 in the flowchart shows 70 sequences of computer-executed instructions performed in the D data processing system to compile the database of representative flow facet examples that were used in connection with the determination of the flow system according to Figure 8. During step 72) Quantities are determined according to the properties The physical phases of the fluid phases above. From these properties" the range of Reynolds numbers is also determined during step 72 for the different flow designs. Noting the parts of the bitumen, water and gas. In step 74 the test matrix is determined where the flow of the bitumen, water and gas varies according to the EIR The Reynolds number is noted for each.During step 75, the physical properties are determined for each 0 point specified in the test matrix, the relevant mixtures of oil, water, and gas flow in the flow loop, and the total flow measured until the Reynolds number is achieved. Step 76 involves defining the test matrix for the flow loop and adjusting the various flows accordingly Reynolds number and phase fractions correspond to the test matrix specified for the well. During Step 78 the scale M' or M' is formed for each of the points of the GURY matrix) and is preferred ; Taking into account 5 moving through the test matrix in as many different directions as possible at random to avoid systemic lag effects. by completing step 78; A coherent set of parts or a database is formed from the m' or m' measurements representing the multiphase flow conditions in a well. The mean flow face is now determined in the manner described above, using the following relation and taking 'm' as an example: my - > 11(q) AM") = 7 Yq M 20 where there are the total measurements of the single flow loop 'q' and parameter 'q' represents the measure 0107 for Q -< 1 to Q".

— 6 1 — Stream image matrices are now defined; " bad " ; (which represent deviations from the mean) using the following relationship: CD — MD _ (M") and so on ¢ A dataset of flow images is collected and stored as 8 sae flow system data and is available to identify flow faces iegen during step 82 of flow faces For flow systems Determine the faces of the flow: Assuming that "m" represents the processed data matrix; The dimension of the flow image matrix 0 0 is s x (n-1). By taking each row and its sequence or association Tae as a group, each CD can be considered as a vector in the s(n-1) dimensional space which using the invention The current is called the flowspace.According to the present invention, what is defined as the flow-faces image ign 0 which is formed during step 82 of the flow-dag data. The flow-faces eigen corresponds to the values of eigen when processing the matrix and represents the stream-image data organized within the matrix image which are described herein. The iGen flow faces are a set of base vectors orthogonal in the flow space that can be added in a linear fashion to construct any flow face image. Thus, the present invention provides arrangement of the stream data in a manner analogous to the inputs for known virtual discrimination 5 techniques. dda can occur (JU) by using one of the processing methods described for example in Sirovich and M.

Kirby, "Low—a dimensional procedure for the characterization of human faces," J.

Opt. (Soc.

Am A, 4 (1987), p. 519-524 previously cited by M.

Turk and A. Pentland, “Eigenfaces for Recognition” J.

Cog.

Neuroscience, 3(1) (1991), p. 71-86; or M.

Turk and A.

Pentland, “Face Recognition Using 20 Eigenfaces”, Proc.

IEEE Conf. on Comp.

Vision and Patt.

Recog., (1991) p. 586-591 It should be understood that Lead can use the AT order of data processing techniques to tag images.

Placing the stream data during step 82 in the manner described produces a first Ign interface vector so that the Euclidean intrinsic product sum between the Ign interface vector and 0091 summed over all values of 0 and scaled with a factor of 1/Q maximum scalar dad. The Optical Society of Americas presents in an article by Sirovich previously mentioned, for example;

A description of how to use this method to sequentially determine each of the faces of the eigen flow (eigen faces). Determine the eigen k values once the eigen flow faces have been defined during step 82; Each of the flow faces is processed during step 84 to define a set of eigen values that can be used to minimize the error between the flow face approximation and the flow face itself. The IGN values generated during step 84 according to the present invention are indicated as the IGN flow images. IGN values represent compressed types

0 as a set of scalar quantities that represent characteristic values of the faces of the eigen fluxes in the matrices. The processing in Step 84 uses the appropriate declining algorithm; like; That which is described in “Numerical Recipes in C: The Art of Scientific Computing”, Cambridge University Press, 2nd Edition (1992), W.H.

Press, S.A. W.T.

Vetterling, B.P.

Flannery, McCam, I© 1. During this minimum depression; is being conducted

5 Subtract the matrix between the close-up image and the real image to produce the difference matrix. Each element of the WAY matrix is squared and all elements of the matrix are summed to provide a scalar dad that increases as the two images become significantly different. A sum of zero over this quantity indicates an identical picture. The error amount is specified as "E". during processing drop to the minimum step 84; Tas processing is for the first eigen flow face only

0 Using the drop-to-minimization processing that was described above is performed; And search for the lowest value of the amount of error "E" where the value of Eigen varies. after that; A second eigen flow face is added and included to be cumulatively reduced to a minimum; And search for the lowest value of the error magnitude "E" where the first two values of Eigen vary in two dimensional spaces. The minimization processing is completed recursively during the "ki" number of these processing steps for the facets

5 eigen "K" flow and eigen AT flow face added for those faces prone to downgrade processing

— 8 1 — 1 term for the lowest described 0 Del during each Ki step of the “facets of the Eigen flow so” the search is performed for the minimum value of Waal 'ie 1 where the first Eigen 'Ki' values vary in a space Dimension AS The required number of eigen flow faces depends on the variance of the flow regimes that require inspection. Based on preliminary analyzes and studies; It is expected that the "ki" number of about 10 steps should provide a reasonable approximation to the final flow face. An important feature resulting from the Ign values offers potential telemetry advantages. using the Eigen value method of the present invention; Using the expected number of "ki" of about 10 steps to decrease to the minimum; It becomes possible to take a data set having points (N-1)s ; Each using an 8-bit representation and condensing it to 10 real numbers. In terms of data compression for a 16 0 channel system using 1000 samples per channel and an 8-bit data point representation; This equates to a memory space of 120,000 bits representing the stream (about 117K). Using the Ign value method described here the mode can be represented for the same polyphasic stream in 320 bits using a single-precision (about 0.3 kbit) free point representation. This means that the data is compressed by a factor of 400 or so without losing important information. 5 on dag select; These data compression techniques independent of the default Shae classification method can be telemetry of low bandwidth in boreholes to transmit the detailed information about the flow mode. Jal) in the flow space: depending on the determination of the faces of the flow eigen during step 84; which are specific to the range of the test matrix specified in Step 74) The present invention provides the ability to represent 0 during Step 86 multiphase flow data on the basis of eigen values alone in the dimensional flow space 'ki'. The representations are provided by means of displays formed using a system Data processing “d n. Depending on the results obtained when testing” such as those shown in Figures 14 and 4b, it is evident that the flow measurement data group together according to the different type of flow system that

has been measured. Thus, it is expected that there will be qualitatively different flow designs where the rates of each phase vary. Research on polyphasic flow points to it as being the case. See for example C.E.

Brennen, “Fundamentals of Multiphase Flows”, Cambridge “till University Press, (2005) for its expected compilation; A virtual marking-through-step 88 technique according to the present invention can now be used to help classify different flow system scenarios. Virtual Classification of Flow Rates or Flow System: Virtual classification processing is a well-known computer processing technique. Representative descriptions of a hypothetical classification are shown; For example, in M. “Noise Tolerant Fine Tuning Algorithm for the Naive Bayesian Learning ‎Algorithm”, J. of King Saud Univ. — Comp. and Inf.

Sci. 26, (2014) p. 0 K.

El Hindi, 237-246; and “Pattern Recognition and Machine Learning”, Springer, Berlin, (2006), C.M.

Bishop according to the present invention; A default classification is used throughout Step 88 to classify systems based on a specific flow phenomenon. This is done so that flow system images are more readily available and accessible for examination and analysis. According to 5 of all's invention shal defaults classification of flow system data depending on any of the flow rates of the flow systems. These two methods appear to be suitable for a multiphase measurement system that has been used on pyrogenic scenarios. It should be understood that shal Load is a default classification based on other flow phenomena as well. as a function of oil flows; water; ally The polyphase flow system is known for putting in different 0 flow patterns. This is described for example” in the paper “Fundamentals of (Brennen (Multiphase Flows”) mentioned earlier. In an easy way; Figures 13 and 3b show how the differences qualitatively between liquid flow and bubble flow are visible visually; Wish) By eye When the flow system database is initially created using the test conditions in the flow loop as described above, a test matrix is designed during step 74 that produces the BASRA dataset 5 The flow system database Next, the flow system database is used Stored flow when processing

Data to determine the flow regime obtained in the flow channels of interest using unknown flow regimes. For each of the obtained cle datasets the present invention allows the flow system of the flow channels to be examined individually. using the present invention; The test matrix of flow system identification data stored in the flow system database is representative of known flow systems. This data is used as probabilities

a priori of known flow regimes that are used as a priori probabilities and distributions of the flow faces. The a priori probabilities represent what is known as the class pattern of a hypothetical classification. By using the a priori probabilities of creating predicted groups; in a process known as checkboxing; For similar ranges of types of flow systems. after that; A default classification during step 88 is performed on the system data

0 flows from unknown flow regimes obtained by transducer matrix 'U.' As described above, the present invention provides the ability to represent flow data from an ultrasonic matrix into low-dimensional de sane of 'ki' variables that can be used To determine the flow. Once the data is mapped in a water way, it is possible to perform a default classification on the measurements. Classification against the flow system: Given a default classification method, the data from each system

5 flows by producing a subset of dimensional vectors "ki" in the flow space. for each flow system; They are each boxed together and a default marking is used during Step 88. A default classification from Step 88 produces confidence levels for the outputs based on the measured data given the probability that the flow is in a given flow regime. Thus a hypothetical classification is used to provide a basis for analysis and identification; during step 88; Whether it is appropriate to use select

0 the flow regime performed during step 88 or whether processing should be performed to specify another flow regime eg; A reflex tomography of the type described in a related pending application pre-identified serial number 595689/14. Ladle reflection tomography could be for example; Regarding the wet gas flow system.

Classification against flow rate: The present invention also relates to a hypothetical classification against flow rate during step 88. In doing so; Care is taken to measure loop flow results using numbers other than absolute flow. This is done so that the flow loop results are plotted relative to the blister flow positions.

5 The rating against the flow rate is comparable to the rating against the flow system. Nevertheless; rather than being put in a box by looking at the flow system; Putting in a box is corresponding to increasing the flow rate; whether overall or by looking at the individual phases. For example (Jal) the flow values specified in the total number of quantitative steps could be equal to Jwater «Joi 985l which could result in a total number of Rs to put into the bins:

Jwater Jgas 10 zo[ = R and so on; virtual education takes place; The output from the processing provides a level of confidence or probability for each value of Jgas (Jwater) that can be represented as a point in a three-dimensional space. Thus, the output of a virtual classification processing can be completed according to the flow rate, whether linear, binary linear, or through Other methods between quantitative values to form a probability field in 3D space 5 provides the maximum value of probability an estimate of the flow of oil, water, and gas This processing also results in a U-matrix based flow measurement from ultrasonic transceivers as described above Considering Fig. 1. Next, the flow system selected during the processing step of a default classification 88 is stored as the flow system ed during step 90 in the memory of the data processing system D. After the flow system selected 0 and the data is also available for viewing and evaluation by Engineers and Analysts Data processing system As shown in Figure 9, the data dallas system of the present invention includes a computer 100 with a processor 102 and memory 4 associated with the processor 102 to store operating instructions, control information and bottom registers. The data inside it. The computer can be 100 (at will; Ble for the Linux group; such as;

available from HP or another source; multi-core processor with nodes; fie those available from IBM (Intel) or Advanced Micro Devices (AMD) A mainframe computer of any conventional type of suitable processing power, such as those available from IBM or from a source CAT 5 It should be noted that other digital processors, such as personal computers, can be used on

PC image; A computer (sane or Beal) a programmable or other suitable programmable digital data processing. The computer 100 has a user interface 106 and an output display 108 to display the output data or registers according to the present invention to form an image of a stream system for a stream

0 - polyphase in channels based on flow system data from 'U' or 'M' transformer arrays. Output display includes 108 components"; Image graphs; data sheets; graphs; datagrams; and the like as logs or output images. The user interface 106 to PC 100 also includes a means for user input

An appropriate I/O controller 110 to provide access for a user to control or access information and database records and computer operation 100. The I/O controller 110 can also receive stream data measurements obtained during data acquisition in the manner described above . The D data processing system also includes a database 112 that is stored in a water memory that can represent internal memory 104; or memory

0 external networked or non—networked as all indicated at 114 on companion database server 116. As noted above; Database 112 also includes the test flow system database that was formed during the processing sequence 70 shown in Figure 7.

The D data processing system includes program code 118 stored in non-transitory memory 000-transitory memory 104 for PC 100. Program code 8 according to Jal's invention is in the form of computer-operable instructions causing data processor 102 modulates stream system images for polyphase stream into channels; As explained above. It should be noted that program code 8 can be in the form of microcode; programs; conventional systems; or symbolic computer-executable languages that provide a specific set of structured operations that control the functional aspect of a D-processing system and direct its operation. Program code instructions 118 can be stored in non-transitional form in memory 104 of the computer 100; or on a computer diskette 0; Magnetic tape; a traditional hard disk drive; Electronic read-only memory “optical storage device” or other appropriate data storage device with a non-transitory computer usable medium stored on it. Code 5 Program 118 Lad can be embedded on a data storage device such as; server 114) on a non-transitory computer readable medium, as shown. From the above, it can be seen that multiphase flow measurements using tomography can be Accurate when there are high levels of free gas 0 within the system that can cause multiple reflections.Tmography generally relies on transmission other than more reflections dada can result from the presence of liquid gas bubbles or Large amounts of free gas combination and multiple reflections can cause confusion [sale] algorithms build direct tomography.Using the present invention, the same data obtained from the tomographic measurement 5 are organized in some way so that they can be fed into virtual discrimination algorithms that use education

automated; and default classification techniques. using the present invention; All data within the measurement are used as corresponding only to portions that can be used to reconstruct the tomography. The enrichment of the received data in complex flow inversions in gas-liquid flows hampers the performance of the regular tomographic reconstruction algorithms; But by using the present invention" shal

This behavior is useful.

The present invention provides increased performance over known tomography techniques. This is particularly the case in wet gas or bubbly flow systems 5 . The present invention may also be of benefit to existing systems by helping to determine the type of flow reconstruction processing to be used.

0 The present invention is sufficiently described so that an average knowledge in this respect can be reproduced and the results stated in the present invention can be reproduced here. Nevertheless; Any person skilled in the field of technology can perform; The subject matter of the present invention is subject to the implementation of modifications not described in the application (lia) to apply these modifications to a specific structure; or in a manufacturing process thereof; the subject matter to be protected requires the following claims; these structures should be covered within the scope of

5 present invention. It should be noted and understood that there are improvements and modifications to be made in accordance with the present invention which are described in detail above without departing from the scope or intent of the present invention which is described in the accompanying claims.

Claims (1)

Protection Elements 1 - A device for determining the flow regime of a three phase flow of oil, water and gas in zl on ultrasonic energy pulses received in a set of transmitting and receiving Seal after passing through the flow three phase flow in the channel; Includes: 5 sets of ultrasonic transceivers installed in an array around the perimeter of the conduit ¢periphery of the conduit 1 set of ultrasonic transceivers AL for individually operating in turn To be sent as Ultrasonic energy of a transducer energy sent to travel through the fluid in the conduit; 0 set of seal ultrasonic transceiver installed in the array around the periphery of the conduit when inoperable for transmitting it is operable for receiving energy in the form of energy pulses ultrasonic After traveling through the fluid in the channel from the transmitting transducer; And an operable data processing system to determine the flow system of 5 Soul three phase flow of oil, water and gas 985 in the channel; It includes: (1) a memory that includes a database that stores the flow system data of the three-phase flow of oil, water and gas as a set of data matrixes that represent systems Known test flows for Oil, Water 0, Gas 985 for a range of in-duct test flow systems; (2) Memory /01607017 also includes stored program instructions to control the operation of the data processing system to perform flow system selection for the three-phase flow phase flow in the channel; (3) A operable signal processing module to perform signal dallas 5. Initialization and coordination of the received ultrasonic energy pulses; (iv) an operable processor under the control of stored program instructions for modeling measurements of the attenuation and transmission periods of ultrasonic energy pulses received between an individual set of Seal transceivers and the rest of the set of transmitters and receivers as a result of travel through the fluid in the channel; (5) The processor is also operable under the control of stored program instructions to regulate measurements formed for the attenuation and travel times of ultrasonic energy pulses received in actual data matrixes indicative of actual oil flow systems oil, water and gas 985 in the canal; (6) The processor is also operable under the control of stored program instructions to form correspondence measurements between formed measurements organized into actual data matrixes and individual groups of data matrixes representing well-known test flow systems for Oil. Water and Gas 985 for Test Flow System Kit; (vii) the processor is also operable under the control of stored program instructions to determine the flow regime of the actual flow conditions of the three-phase flow of oil, water, gas 985 in the conduit based on symmetric measurements; and (viii) an operable display under the control Stored program instructions for presenting flow system specified for actual flow conditions for evaluation and analysis 2 - Gay device for claim 1; also includes data processing system 0 5715160 which includes: memory also includes flow system Store of the actual flow conditions determined using the processor 3 - Device By of claim 1 Cua that the processor when operable under 5 is controlled by the stored program instruction to form correspondence between the structured formed measurements of the data matrices The actual matrixes and individual groups of the test condition data matrixes Runable to perform Bayesian pb classification of data matrixes under actual flow conditions. 4 - The device according to Claim 3 Gus that the Bayes classification for the formation of symmetry between the structured formed measurements of the actual data matrixes and the individual groups of the set of test condition data matrices includes the Bayes classification depending on the flow systems. 5 - Device in accordance with claim 3 wherein the ph rating for modulation Aa symmetry measurements organized for actual data matrixes and individual ensembles of test conditions data matrixes includes Bayes rating depending on flow rates 6 - system dallas data processing system to determine the flow regimes of a polyphase flow in ald) based on the transmission of energy through the fluid in the channel from a transceiver transmitting in an array of transceivers around the periphery of the channel the conduit 5 to a set of transceivers that receive in a seal array transceiver in the array The data processing system includes: (a) 5505 memory including database The database stores the flow system data of the three phase flow of oil, water and gas 985 as a set of 0 data matrixes that represent the well-known test flow systems for Oil, water and gas 985 for a group of other flow systems tape in the channel; (b) the memory also includes stored malin instructions to control the operation of the data processing system to make a determination of the flow regime for the three phase flow in the channel; (c) an operable signal processing module to initially signal dallas and coordinate the ultrasonic energy pulses received through the fluid in the channel; (d) an operable processor under the control of stored program instructions for modeling measurements of the attenuation and travel times of ultrasonic energy pulses received between an individual set of Seal transmitters and receivers and the rest of the set of transmitters and receivers as a result of travel through the fluid in the channel; (e) The processor is also operable under the control of stored program instructions to regulate the measurements formed for the attenuation and travel times of ultrasonic energy pulses 0 received in actual data matrixes dlls on systems actual flow of oil, water and gas 985 in the canal; (f) The processor is also operable under the control of the instructions of the stored program to form correspondence measurements between the formed measurements organized in the actual data matrixes and the individual groups of the data matrixes representing the well-known test flow 5 systems of oil and water and Gas 985 in the channels for the Test Flow Systems group; (g) the processor is also operable under the control of stored program instructions to determine the flow regime of the actual flow conditions of the three-phase flow of oil, water, and gas in the conduit based on symmetric measurements; and (h) an operable display under stored program instruction control to present the specified flow system 0 to the actual flow conditions for evaluation and analysis. 7 - The data processing system “dallas” according to the protection element 6 “also includes: a memory that also includes a flow system that stores the actual flow conditions that were determined 5 using the processor. 8 - Data dallas system Data processing system according to protection element 6 where the processor is operable Shall Bayes classification of data matrixes Actual flow conditions in the formation of symmetry measurements of the data of the actual flow conditions to a set of data matrices test conditions. 9- The data processing system, according to Claim 8, where the Bayesian classification for the formation of symmetry measurements for the data of the actual flow conditions to the set of data matrices of the test conditions depends on the flow systems. 0 10 - The data processing system according to claim 8) whereby the Bayesian classification for the formation of symmetry measurements of the data of the actual flow conditions to the set of test condition data matrices depends on the flow rates. 1 - the device according to claim 1; Where data includes a flow system known as triple flow 5 Phases Oil all, water and gas 985 on: Three-phase flow rates in the test flow regimen group. 2 - the device according to claim 1; Where lily includes the well-known three-phase flow system for the Oil, Water and Gas 985: 0 three-phase viscosity and density in the group of test flow systems. 3 - device according to claim 1; Known flow system data for the oil, water, and gas 985 include: a mixture of viscosities and densities for the three phases of the three phase oil all flow 5 and water and gas 985. — 0 3 — 4 - Gag data processing system dallas for claim 6 where known flow system data for the Oil, Water, Gas 985 three-phase flow system includes: Three-phase flow rates in the group of test flow systems. - Data dallas system dy data processing system for Claim 6 where the known flow system data for the three-phase flow of Oil, Water and Gas 985 includes: viscosity and density of the three phases in the group of test flow systems. 6 - Data dallas system Gig data processing system for claim 6 where 0 The data of the known flow system of the three-phase flow of oil, water and gas 985 includes: a mixture of viscous materials and densities 0605186 for the three phases of the three-phase flow Phase flow for oil and water .gas lly JPY AAM AAM L nd 3 FCA of a RE 1 QBA AAA 2 & FF of .TAA 83 aa x Fa a wos go l kr and i 1 The >< from N B A tT x EYEE A + LAB A B .: A L A ps 0 A 3 1 1 Bm .| L AG CV SA NA B: Lt Revealed Se ON Non { Ea A A A A A A A A A Laa A Aak Ee A eT A 0 The Collection x > » Squirt Ny EEA a Sr A L D NO TN SE SEA en A AN L L L R TH A A A A A A A A A A A A A A A Squirt A A A S A A A A A A A A Squirt § 3 p eat Wh; Kirat SAR ECR iy 1 1 Enea 5 TS | ا ا ا ا ا ا ا ا ا ا % RITE SR Ada Reply SO K 3 L L L A I D A A L A No M B "Al Ahadi Q J A A L 5 ¥ 4 : 8 For 0 8 5 2 Anad Ta aR 0 A FN TET Na Aa Ny Ty Give me w ey ke 3 A A ar I CaN We oT TRARY EY yt Te i LENT NTR Sy “ i ST C SL A Yo LA i 3 Tami A Yo LA i Alif Te N : as sb br l v a except the L F a NX ov B Yo NS So Vo 1 MB = 8 5 EE 4 \ J > x re A CS NON A \ / J I ANY EA TONY LSE 4 Gard Soo E : oo a >” ¥ A N ad a Noid ra Sb re A AY hy . A © 1 Ma 'Ya Yijia' B 4 4 8 aaa 0 ko z & And . Ee. bjn -- > "1 La y Be aE 0 aaa 0 form x oN ss --0-- . yr 0" = fo ; Ba Jad LIN 5 hy eh 1 Se ® Ny oh > A + B"‎ 2 8 Lad w 1 oF Fag &® LTR SY Fw Be ®ha 2% aa mag seal FF 8 I 1 for a magazine Xi TE A EO PIER B Melody Lokma Ri : : > ast & =. IN a YE ia yl 5 The My AE LY ® i 2 book Pe TR 0 REEL B L MD D YAD Eb Sin Les 1 DA : att radi Fe ou fairy | & $e .wet * § 8 * 8 % ® Ba 13 Sp * oy x =e LRTI HR Ge Z Woy ir H 3 of i º 1 º ' 1 um and um Hy FA dus i ANE 1 a LW RENE § ERR RY a il ] hy 0 8 8 d Ota A L SEN JLT TREE CRT Se first entry Boo AA Ak form “B”. — 5 3 — lv 3 as, lvs ’ b ye | : , os 4 ld form + : hy 3 9 3 ak 0 1 : ; 1 1 | ami Pb 3 a mC { 1 E 3 1 \ s tbg | we i - \ 3 - m a 5 8 3 hayer: | 8 i. : Be i { RE N : : \ Veo oF 1 1 | 1 1 : : 8 i | =< | jEI. 0 3 mi for : } bd i Sn - . § {oe : |3 but 1 ET 3 SE : ) 5 : SES | 1 Magh Maj Begum 3 a i | So ] Ty 3 d Atg La = TT 2 a 3 . that 1 : = & * ga g Sued Authority for Intellectual Property RE .¥ + \ A 0 § 8 Ss o + < M SNE A J > E K J TE I UN BE Ca a a a ww > Ld Ed H Ed - 2 Ld, provided that the annual financial consideration is paid for the patent and that it is not null and void for violating any of the provisions of the patent system, layout designs of integrated circuits, plant varieties and industrial designs, or its implementing regulations. Ad Issued by + bb 0.b The Saudi Authority for Intellectual Property > > > This is PO Box 1011 .| for ria 1*1 v= ; Kingdom | Arabic | For Saudi Arabia, SAIP@SAIP.GOV.SA